[Bug c++/18071] New: -Winline does not respect -fno-default-inline

[Bug c++/18071] New: -Winline does not respect -fno-default-inline

[markus at oberhumer dot com]

-Winline does not respect -fno-default-inline (this is split out of bug17115,
and a possible 3.3 regression).

Consider this code:

struct Foo {
            int a(int r) { return r & 1; }
    virtual int b(int r) { return r & 2; }
    static  int c(int r) { return r & 4; }
};

int bar(int r) {
    Foo f; int k = 0;
    k |= f.a(r); k |= f.b(r); k |= f.a(r);
    return k;
}

> g++-3.3 -c -O2 -fno-default-inline -Winline test.cpp
> g++-3.4 -c -O2 -fno-default-inline -Winline test.cpp
test.cpp: In function `int bar(int)':
test.cpp:6: warning: inlining failed in call to 'int Foo::a(int)': --param
max-inline-insns-single limit reached
test.cpp:13: warning: called from here
test.cpp:7: warning: inlining failed in call to 'virtual int Foo::b(int)':
--param max-inline-insns-single limit reached
test.cpp:13: warning: called from here
test.cpp:6: warning: inlining failed in call to 'int Foo::a(int)': --param
max-inline-insns-single limit reached
test.cpp:13: warning: called from here
test.cpp:6: warning: inlining failed in call to 'int Foo::a(int)': --param
max-inline-insns-single limit reached
test.cpp:13: warning: called from here
test.cpp:7: warning: inlining failed in call to 'virtual int Foo::b(int)':
--param max-inline-insns-single limit reached
test.cpp:13: warning: called from here
test.cpp:6: warning: inlining failed in call to 'int Foo::a(int)': --param
max-inline-insns-single limit reached
test.cpp:13: warning: called from here


Futhermore, the warnings disappear if you use -O3 instead of -O2, so -O3 seems
to silently override -fno-default-inline (at least a documentation bug).


Full test program and log file will follow as attachment.

-- 
           Summary: -Winline does not respect -fno-default-inline
           Product: gcc
           Version: 3.4.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: markus at oberhumer dot com
                CC: gcc-bugs at gcc dot gnu dot org
  GCC host triplet: i386-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug c++/18071] -Winline does not respect -fno-default-inline

[markus at oberhumer dot com]

------- Additional Comments From markus at oberhumer dot com  2004-10-20 05:49 -------
Created an attachment (id=7383)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=7383&action=view)
Test program, and log of compilation with various gcc versions.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug c++/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-20 14:00 -------
Confirmed, a 3.4 and 4.0 regression.

But note -O3 enables -finline-functions which enables inlining for all functions so that is not a 
documentation problem at all.

: Search converges between 2004-01-01-trunk (#437) and 2004-01-17-trunk (#438).

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
           Keywords|                            |diagnostic
   Last reconfirmed|0000-00-00 00:00:00         |2004-10-20 14:00:23
               date|                            |
            Summary|-Winline does not respect - |[3.4/4.0 Regression] -
                   |fno-default-inline          |Winline does not respect -
                   |                            |fno-default-inline
   Target Milestone|---                         |3.4.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-10-20 21:52 -------
Mine.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |giovannibajo at libero dot
                   |dot org                     |it
             Status|NEW                         |ASSIGNED
          Component|c++                         |middle-end


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2004-11-01 00:46 -------
Postponed until GCC 3.4.4.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.4.3                       |3.4.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-11-23 13:33 -------
I eventually managed to test the patch, but there are testsuite failures:

FAIL: gcc.dg/debug/dwarf2/dwarf-die7.c scan-assembler 1.*DW_AT_inline
FAIL: gcc.dg/winline-2.c  (test for warnings, line 4)
FAIL: gcc.dg/winline-2.c  (test for warnings, line 7)
FAIL: gcc.dg/winline-4.c  (test for warnings, line 4)
FAIL: gcc.dg/winline-4.c  (test for warnings, line 7)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-11-23 13:41 -------
FAIL: gcc.dg/debug/dwarf2/dwarf-die7.c scan-assembler 1.*DW_AT_inline
Was there before your patch if you are testing on 4.0.0.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2004-11-24 00:34 -------
Giovanni, your patch actually looks correct to me. 
 
It may be just exposing a latent bug in the C front end.  I'm not 
sure but I'd expect that if a function is declared inline as in 
testsuite/gcc.dg/winline-2.c: 
 
/* { dg-do compile } */ 
/* { dg-options "-Winline -O2" } */ 
 
inline int q(void);              /* { dg-warning "body not available" "" } */ 
inline int t(void) 
{ 
        return q();              /* { dg-warning "called from here" "" } */ 
} 
 
then q should be DECL_DECLARED_INLINE - but it looks like it's not, 
or you shouldn't see those new FAILs. 
 
 
 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-11-24 01:54 -------
JSM, do you agree with Steven's analysys? It looks like DECL_DECLARED_INLINE 
should just mean there, so the C frontend might be wrong in this regard.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jsm28 at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[joseph at codesourcery dot com]

------- Additional Comments From joseph at codesourcery dot com  2004-11-24 16:57 -------
Subject: Re:  [3.4/4.0 Regression] -Winline does not
 respect -fno-default-inline

On Wed, 24 Nov 2004, giovannibajo at libero dot it wrote:

> JSM, do you agree with Steven's analysys? It looks like DECL_DECLARED_INLINE 
> should just mean there, so the C frontend might be wrong in this regard.

I agree the C front end should set DECL_DECLARED_INLINE when a function is 
declared inline.  Before looking for a problem in that regard, make sure 
that (a) DECL_DECLARED_INLINE is what is checked for the diagnostics in 
question, (b) it is indeed not set for the relevant decl.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-11-25 00:20 -------
The problem appears to be the other way round. The C frontend mark the 
declaration as DECL_DECLARED_INLINE but not as DECL_INLINE because the body is 
not available. The relevant portion of code is this:


	else if (declspecs->inline_p)
	  {
	    /* Record that the function is declared `inline'.  */
	    DECL_DECLARED_INLINE_P (decl) = 1;

	    /* Do not mark bare declarations as DECL_INLINE.  Doing so
	       in the presence of multiple declarations can result in
	       the abstract origin pointing between the declarations,
	       which will confuse dwarf2out.  */
	    if (initialized)
	      {
		DECL_INLINE (decl) = 1;
		if (storage_class == csc_extern)
		  current_extern_inline = 1;
	      }


The C++ frontend instead does not have this issue: by default, all declarations 
marked by DECL_DECLARED_INLINE are also marked with DECL_INLINE. In fact, the 
testcase works even after my patch when compiled with cc1plus.

The code has been there since this patch: http://gcc.gnu.org/ml/gcc-cvs/2001-
12/msg00833.html. It fixed PR 5163. I tried removing the if(initialized) and 
always set DECL_INLINE even for functions with no bodies, and the testcase for 
the bug still succeeds. Maybe the weirdness in dwarf2out has been fixed since 
then.

RTH, do you think it makes sense to remove that check and set DECL_INLINE even 
for functions with no bodies in the C frontend? The C++ FE does that.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rth at gcc dot gnu dot org,
                   |                            |stevenb at suse dot de


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-11-30 11:25 -------
Patch posted here, waiting for review:
http://gcc.gnu.org/ml/gcc-patches/2004-11/msg02664.html

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[markus at oberhumer dot com]

------- Additional Comments From markus at oberhumer dot com  2004-11-30 19:54 -------
Just a short note that the testcase has a typo and should read

   ...
   k |= f.a(r); k |= f.b(r); k |= f.c(r);
   ...


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-12-01 07:15 -------
Subject: Re:  [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

markus at oberhumer dot com <gcc-bugzilla@gcc.gnu.org> wrote:

> Just a short note that the testcase has a
> typo and should read
>
>    ...
>    k |= f.a(r); k |= f.b(r); k |= f.c(r);
>    ...


Right thanks for noticing. Luckily, this does not expose a bug in my patch,
which is still valid. I'll make sure to commit the correct testcase when I get
to it.

Giovanni Bajo



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0 Regression] -Winline does not respect -fno-default-inline

[markus at oberhumer dot com]

------- Additional Comments From markus at oberhumer dot com  2005-01-21 02:10 -------
What is the status of the latest patch from Giovanni ?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0/4.1 Regression] -Winline does not respect -fno-default-inline

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2005-04-29 18:30 -------
Removing rejects-valid marker; spurious warnings do not count as rejects-valid.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|rejects-valid               |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0/4.1 Regression] -Winline does not respect -fno-default-inline

[mmitchel at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.4.4                       |3.4.5


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0/4.1 Regression] -Winline does not respect -fno-default-inline

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-07-22 21:13 -------
Moving to 4.0.2 pre Mark.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.4.5                       |4.0.2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071

[Bug middle-end/18071] [3.4/4.0/4.1 Regression] -Winline does not respect -fno-default-inline

[mmitchel at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.0.2                       |4.0.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18071